Articles |
Consulting the Development-Displaced Regarding their Resettlement: Is there a Way?
University of Liverpool Management School, Liverpool, UK s.garikipati{at}liverpool.ac.uk
Involuntary resettlement of people displaced by large dams, as by several other development projects, is largely considered a failure. Increasingly, researchers have shifted focus from trying to improve welfare outcomes for the affected to questioning the involuntariness of resettlement schemes. The idea of voluntary resettlement, under which the displaced are consulted regarding their resettlement, is often suggested as a preferred alternative. The problem of ‘incentive incompatibility’, however, suggests that, if asked for an evaluation of their losses, the affected have an incentive to exaggerate. In this paper the author explores the possibility of using the contingent valuation method—a method often used to make decisions about environmental issues—to consult the affected without capitulating to this problem. This experiment is carried out in nine villages affected by the construction of the Sardar Sarovar Project in the Narmada Valley.
Ms received August 2004; revised MS received April 2005
1. Several other groups, such as those affected by the canal network, secondary displacement, and the downstream effect, are left out of the compensatory net and their crusade against this injustice continues (for details and estimates, see Morse and Berger 1992; Fisher 1995).
2. In his seminal model of ‘Risk and Reconstruction’, Cernea (1995) cautions against eight key risks that may be responsible for impoverishing a displaced population: landlessness, joblessness, homelessness, marginalization, food insecurity, loss of access to common property resources, increased morbidity, and community disarticulation.
3. The names of the survey villages used in this study are fictitious. The names Gaman and Mapali are taken from Hakim (1995).
4. When referring to ‘tribal’ people in the Indian context it is more appropriate to use the term ‘adivasi’, which literally means ‘original dweller’, and is the term used by the people to refer to themselves (Hardiman 1987). The use of ‘tribal’ is also problematic because of the porosity of the boundary between ‘caste’ and ‘tribe’, which have a parallel existence in India. For reasons of wider recognition, however, I have used the term ‘tribal’ to refer to the ‘adivasi’.
5. Phalias share a common water source and are usually inhabited by members of the same tribe. At times these were more than 10 kilometres apart and hence it was not uncommon to find women of one phalia who had not visited other phalias for years. Most labour sharing activities were also contained within a phalia.
6. Fifty-nine families registered at the R-sites had actually moved out (this constitutes 32.8 per cent of families still living at these sites). Most of them were found to be living in the PS-villages but were not officially recognized to be doing so (A. B. Chaudhary, Resettlement Official of the Sardar Sarovar Narmada Nigam (SSNN), personal communication, 26 May, 1998).
7. Pearce (1999) was the first to recommend that CVM should be used to evaluate preferences for displacement packages, although the first stage of the main survey reported here was carried out prior to this recommendation.
8. Although economic theory contends that WTA and WTP differ only by an income effect, CVM surveys have empirically established that WTA is almost always higher than WTP, often substantially higher (Hanemann 1991; Amiran and Hagen 2003). Any attempt to study the relationship between WTA and WTP was abandoned very early on in the survey because eliciting WTP to avoid displacement resulted in either ‘protest zeros’ or nonparticipation.
9. It is likely that, if interviewed separately, women may participate with greater vigour and may hold preferences that are different from those of their men. This study, however, did not attempt to interview them separately because it aims to capture what a household's resettlement preferences would be, given an opportunity, and not gender differences in these preferences, although this in itself is an important issue (see, for instance, Sparkes 1997).
10. One of the drawbacks of the close-ended CVM approach is that it says little about individuals' WTA/WTP (Hanemann and Kanninen 1999). Hanemann (1985) proposes the use of a follow-up question to improve the efficiency of this approach. This is known as the double-bounded model and involves asking the respondent for a second bid. Although second bids were not explicitly asked for, some respondents volunteered this information.
11. One popular suggestion was that all should pool their money and use a third of it to buy land, a third to construct houses, and deposit the remaining in a bank for their children. About five per cent of those who rejected the cash package said they would accept the package if offered more money.
12. A binary choice model includes only categorical (‘accept’ and ‘reject’) responses. Excluding noncategorical responses could lead to a sample selection bias. One way of testing for such bias is by plotting the excluded observations against various regressors and checking for any systematic patterns. Such an exercise revealed no patterns, justifying the use of a binary model.
13. While the partial derivative 
is a good approximation for the marginal effect of a continuous variable, for a dummy variable the appropriate measure is the difference of the predicted probability conditional on each of the two categories (0 and 1) (Greene 2003; Roncek 1993). The variance of for dummy variables was computed using the Delta method (Greene 2003), which was in turn used to calculate the t-statistics.
14. Where the CVM measures direct use values (as here), it is suggested that results are verified by comparing the estimates with those obtained through indirect methods (Bishop and Heberlein 1990).
15. Two further results strengthen this conclusion. First, the average land cultivated by the respondents who have accepted this package is 4.2 acres less than that cultivated by those rejecting it. Second, although the marginal effect is negligible, land has a coefficient that is significant and negative.
16. This evidence supports the conceptual framework suggested by Scudder (1991) for evaluating new settlements. He argues that members of communities undergoing relocation react in predictable and broadly similar ways. During the initial period they avert risks by clinging to familiar practices, but as they re-establish themselves socio-economically they begin to take risks. Their attitudes become increasingly individualistic; this may happen because of the breakdown of communal organization and leadership that hitherto restrained individual initiative.